Geoforum 32 (2001) 471482

www.elsevier.com/locate/geoforum

Territorial exclusion and violence: the case of the state of S~o Paulo, Brazil a
Raquel Rolnik a,b
b a Plis Institute, S~o Paulo, Brazil o a Faculdade de Arquitura e Urbanismo, Campinas Catholic University, Brazil

Received 5 November 1999; in revised form 4 October 2000

Abstract If we had to point out one single feature to dene Brazilian cities today, it would be the existence of a dual built environment: a landscape produced by private entrepreneurs and contained within the framework of detailed urban legislation, and another one, three times greater, self-produced by the poor and eternally situated in an intermediate zone between the legal and the illegal. In addition to being an expression of economic and social disparities, this contrast has profound implications on the form and functioning of the cities. The sprawl of the precarious peripheries has lead to an absurd disconnection of poorly urbanized spaces and the city center where jobs, cultural and economic opportunities are concentrated. The eects of this persistent territorial exclusion are devastating and occur in both the peripheries and the city center. The purpose of this paper is to explore the nexus between the precarious and risky urbanization which took place in Brazilian cities and the urban violence that seems to be the most recent and visible face of this model, using the concrete example of dierent cities in the state of S~o Paulo. In order to construct the a links it is rst important to understand how patterns of economic development and population trends have contributed to the generation of risk urbanization and how planning and urban management policies interact with it. 2001 Published by Elsevier Science Ltd.
Keywords: Territorial exclusion; Risky urbanisation; Social exclusion; Brazil

1. Introduction If we had to point out single feature that is pervasive and powerful in Brazilian cities today, it would be the existence of a dual built environment: one landscape produced by private entrepreneurs and contained within the framework of detailed urban legislation, and another, three times greater, that is self-produced by the poor and situated in an intermediate zone between the legal and the illegal. In addition to being an expression of economic and social disparities, this contrast has profound implications for the form and functioning of the cities. The sprawl of what I call ``precarious peripheries''1 has led to a tremendous gap between poor spaces with precarious or inexistent urban infrastructure and the city center where jobs and cultural and economic opportunities are concentrated. The eects of this persistent territorial exclusion are devastating and occur in both the peripheries and the city center.
E-mail address: rrolnik@polis.org.br (R. Rolnik). 0016-7185/01/$ - see front matter 2001 Published by Elsevier Science Ltd. PII: S 0 0 1 6 - 7 1 8 5 ( 0 1 ) 0 0 0 1 7 - 3

In the peripheries, urbanization is incomplete and mostly risky. Risky urbanization is marked by vulnerability: caused by the insecurity of land title ownership to the instability of the terrain and building structure. The lands where these residential markets for the poor develop are generally the most fragile, vulnerable, dangerous and hard to urbanize from an environmental point of view: they are often on steep slopes, river banks and swamps. At times
1 ``Periphery'' in Brazilian cities means a place situated at the outskirts of an existing city, were non-urbanized land is divided into small parcels and sold to poor families by installments. On these parcels, families build their houses themselves, little by little, mostly during the weekends. Sometimes parts of these houses are rented by other poor families before they get access to their own parcels and frequently young couples (recently married sons and daughters) and relatives share the same plot, adding rooms or houses to the original one. These ``loteamentos populares'' (or low income sub-divisions) do not meet the minimum standards (including percentage of public space, width and length of streets, minimum lot size, provision of basic infrastructure) to be approved as developments by local authorities; therefore, they are opened without being registered and recognized as new streets by the city.

472

R. Rolnik / Geoforum 32 (2001) 471482

these areas are ``protected'' by environmental laws and regulations; yet they often have the highest growth in population and land use. The building structures are rarely stable and never really completed (residents are continually adding makeshift additions to their homes). In these non-regulated developments, land ownership is almost never registered in city records and land titles. Residents of the periphery are those most at risk in the city: houses can slide or ood with heavy rains, drainage and sewerage merge in low lands life and health are in permanent danger. People daily lose hours in inecient transportation systems, and live with the discomfort of private houses and public spaces and uncertainty about the future of the neighborhood. But risky urbanization has an impact on the whole city, not just the peripheries. By concentrating assets in a small part of the territory and preventing all citizens from sharing these resources, the best-equipped spaces of the city are constantly menaced by real estate pressure, congestion and robbery. When the erosion caused by deforestation and urbanization of slopes accumulates in low lands, rivers and streams, the whole city suers with oods. When too many people from the peripheries commute to the center, they create excessive trac jams that cut the circulation of the entire city. Territorial exclusion is thus, more than the image of inequality, it is the condemnation of the city as a whole a risky urbanization. The purpose of this paper is to explore the nexus between risky urbanization and the urban violence that seems to be the most recent and visible face of this kind of urbanization, using the concrete example of dierent cities in the state of S~o Paulo.2 In order to construct a the links, it is rst important to understand how patterns of economic development and population trends have contributed to the generation of risky urbanization and how planning and urban management policies interact with it. The empirical base of this paper is a study that was designed to evaluate the impact of land use regulations in the functioning of residential markets in the cities of the state of S~o Paulo with more than 20,000 residents. a The basis of the research was a survey applied to 220 cities3 and was responded to by 118 cities. By using the
2

data from this survey combined with a special crossanalysis of the 1991 Census, we were able to evaluate to what degree planning and land use control instruments which in principle are designed to provide environmentally sound and socially balanced citieshave achieved their goals in the cities of the State of S~o a Paulo.4 The survey explored the planning urban regulation processes in the cities and the timing and conditions in which they were produced and implemented. With this information we organized a ranking of cities according to the existence of dierent land use control regulations, in such a way that the cities were listed from the ``most regulated'' to the ``least regulated''.5 The special cross analysis of the 1991 census was used to construct a ``territorial exclusion'' indicator using information concerning household conditions. Four sets of information were assembled in the matrix (housing conditions, location, infrastructure availability and number of rooms), and they were transformed into dichotomous variables (adequate or inadequate).6 Thus, instead of measuring averages, the indicator measures the percentage of urban residences in a given city excluded from basic living conditions. The ``excluded''
4 The survey was responded to, in each city, both by a representative of the planning sector and of the legal department of the city. Although we could not check the reliability of the data, we can consider it to be high for two reasons: (a) the data were crossed with that from two other similar surveys and (b) the questions concern mostly objective subjects such as the existence or not of certain urban regulation measures. The social- economic data were taken from the 1991 national census, which is considered to be a reliable source. 5 We have considered approved master plans, land use and occupation control regulations, development regulations and other urban legislation, specically linked to environmental protection and/or driven to open access to residential markets for urban poor. The survey makes a distinction between regulations that are only in the statute and the ones that are really applied, and this dierence was considered afterwards, when regulation and territorial exclusion were superposed. Based on the answers from the survey, (points were attributed to each itemof urban regulation existing in the municipality. The eective implementation of urban infrastructure, services or mechanisms corresponded to higher points than those where eorts were merely proposed or in the planing stages. Thus, the existence of a completed Master Plan corresponds to 1 point; one that is approved: 1.5 points; regulated: 2.25 points, and one that is in vigor or being revised: 3 points, and so one, for each item of urban regulation). 6 For instance to evaluate infrastructure, four variables were taken into account water supply, sewerage, waste collection and public lighting. If the residence is connected with the public water system with internal pipes, it is considered adequate; all other systems (ponds, public water faucets, etc.) are inadequate. For sewerage, sealed septic tanks or public systems are considered adequate while all other possibilities are inadequate; for waste collection, waste collected directly or indirectly is adequate; all other solutions (burnt, buried, left in empty lots, thrown in the rivers, lagoons or sea) are inadequate. For public lighting electric systems with meters are adequate while all the rest (electric systems without meters, oil or kerosene) are inadequate. We call the index which results from the crossing of the data above percentage of residence adequacy.

S~o Paulo is the name of a city with 9.8 million inhabitants city. It a is also the name of a metropolitan region with 16 million inhabitants and of one of the most populated and urbanized state in Brazil (35 million people in 625 municipalities). 3 S~o Paulo state has 220 cities with more than 20,000 inhabitants. I a chose cities of this size because the Constitution of 1988 determined that all cities with more than 20,000 people are obliged to implement planning processes and urban land use regulations in their urban areas. The same determination existed before, since 1975, when a federal decree restricted the access of a city to credit and loans (including external cooperation) to expand urban infrastructure and services to those which presented a municipal master plan in which the project was included.

R. Rolnik / Geoforum 32 (2001) 471482

473

territories may be part of inner city slums, illegal subdivisions in the peripheries or other forms of informal housing marked by some form of precariousness of the built environment. The concept of ``territorial exclusion'' was forged in order to overcome the diculties of dealing with traditional indexes of infrastructure coverage and general indicators of housing conditions which do not provide a faithful image of the dierences of living conditions within a given city. By crossing the indicators, we could draw a clearer picture of where urbanization was ``complete'', and where it was precarious or incomplete for some reason. These indicators could also help measure urban social segregation, since these data also can be crossed with family income, income of head of family, race and other economic and social variables. We have chosen the term ``territorial exclusion'' for the obvious purpose of linking it with the concept of social exclusion (Sposati, 1996), rather than poverty or class disparities. This concept which relates the accumulation of various types of deciencies to the lack of social ties has been increasingly used in public policies and can be understood as the denial (or a disrespect for) rights that guarantee each citizen a minimum standard of living, and participation in social and professional institutions and networks. (Castel, 1995; Paugam, 1996). Social exclusion is seen as a way of analyzing how and why individuals and groups fail to have access to or benet from the possibilities oered by societies and economies. The notion of exclusion considers both social rights and material factors. It encompasses not only the lack of access to those goods and services which provide satisfaction of basic needs, but also exclusion from public safety, justice, representation and citizenship. (Rodgers, 1995). Territorial exclusion diers from segregation insofar as it demarcates not only dierences in the quality of the space and its links to social dierences but also means deprivation in terms of access to a basic degree of urban life and opportunities. The hypothesis of this study is that social exclusion in Brazilian cities, not only has a very clear territorial expression, but that territorial exclusion is one of the most powerful causes of social exclusion.7

To best analyze the research ndings, we crossed the data with additional data about the cities including: population growth rates, value added per capita, municipal revenue per capita and percentage of head of residences with income below two minimum wages.8 All the information was placed in a GIS database in order to create a map of territorial exclusion in S~o Paulo a state. To deepen the analysis in the second phase of the study case studies were conducted of three cities chosen from among the 118 cities because they were considered to be representative of the dierent patterns of territorial exclusion found in the rst phase. Each city was considered historically and more critically. Finally, we crossed the data with available indicators of urban violence.9

2. Mapping territorial exclusion in the state of S~o Paulo a Fig. 1 (percentage of residences with an adequate situation) gives us a very clear picture of the regionalization of ``precarious urbanization'': the worst situations of territorial exclusion are found on the extent of the periphery of the S~o Paulo metropolitan region a (Embu-Guacu 1.3%; Aruj 6.26%; Francisco Mora  ato 7.46% and Rio Grande da Serra, Cotia, Emb, u Cajamar and others);10 but the phenomena repeats itself in the peripheral cities of Baixada Santista, North littoral, Campinas, Para Valley and Sorocaba. Precarba ious urbanization is found in exactly the same locations as in the most dynamic and wealthiest regions of S~o a Paulo State, to where large industrial companies have moved since the 1970s. From an urbanistic point of view this region marks the radius of an urban development pattern based on large industrial zones, on transportation systems based on cars, trucks and buses and on the expansion of a precarious housing for the poor, spreading the extent of territorial exclusion. The crossing of per capita production in manufacture with the previous variable reinforces the argument among the 10 cities with the greatest added value per capita11 production in S~o Paulo State, six belong to the a group of the cities with the lowest percentage of resiMinimum wage in Brazil is US$110 a month. Concerning the reliability of the violence dates, we have worked mostly with gures of homicides per capita, taken from the registration of the causa mortis of each one of the state's dead. Those were considered the most trustful data to measure violence, coming from public health authorities that have a very consolidated system of registering death by cause in the State of S~o Paulo. This was the way a we found to avoid distorted data due to fear of violence in certain areas, or any other reason. 10 See Table 1 percentage of residences with an adequate situation (cluster analysis). 11 Added Value is the most important indicator used in Brazil to measure industrial production.
9 8

This could be illustrated by Auyero (1999), writing about a slum in Buenos Aires that ``an invisible wall of economic redundancy, educational exclusion, state abandonment and sustained stigmatization has been erected (separating this slum from the rest of society). The above `cries' synthesize the paramount reality of slum dwellers: a highly oppressive mix of everyday violence and humiliation, state corruption, educational failure, joblessness or extremely precarious occupation attachment, and increasing drug-consumption and trafcking. This blend fosters social and physical insecurity''. In this sense, we could say that territorial exclusion in not just a Brazilian issue, but at least a South American one.

474

R. Rolnik / Geoforum 32 (2001) 471482

Fig. 1. Level of urban regulation (horizontal axis) by level of territorial exclusion (vertical axis).

dences with adequate living conditions. In these cases, companies locate their factories in a small municipality connected by road systems to a potent center (like Monte Mor in the periphery of Campinas, or Mau, at the pea riphery of consolidated ABC;12 or Cubat~o between a the city of S~o Paulo and the port city of Santos). The a location of industrial plants attracts workers while simultaneously contributing to environmental degradation and the creation of illegal housing markets for the poor. These economically vibrant cities are increasingly inhabited by an exclusively poor population (since managers and executives live in the nearby center or in S~o Paulo) and are growing at astonishing levels. The a pattern is a form of economic development, with high levels of income concentration and strong regional inequalities, which leads to territorial exclusion.13 Another pattern, also linked to the presence of nearby wealth is found at the tourist resorts located in the same macro metropolis. Coastal cities like Guaruj a or mountain resorts, like Campos de Jord~o, t into this a pattern. They were developed by real estate investors who oer weekend, second residences for dierent
ABC is a vast region in the Southeast of S~o Paulo Metropolitan a region, with 2.2 million inhabitants and a strong industrial economic base. It held also strong labour unions which started taking part in the country's politics from the end of the 70's on. 13 Telles, 1995 gives us an important contribution in his article ``Structural sources of socioeconomic segregation in Brazilian metropolitan areas. The author concludes in this text that industrialization itself does not increase segregation, in fact it reduces it. But industrialization in Brazil comes together with enormous urbanization and larger cities mean more segregation. Besides, higher income due to industrial wages gives high and high-middle classes money to invest in its self-segregation from lower classes''.
12

market segments (from middle-class apartments to luxurious condominiums and marinas), attracting construction workers and domestic servants and also functioning as dormitory cities for neighboring industrial centers (the case of Guaruj and the Santos port). a In these cases, we have municipalities with the lowest value added per capita production neighboring municipalities with the highest. This pattern of territorial exclusion has a lot to do with income distribution, and mostly to poverty. As we have seen, the worst conditions in the State occur alongside the most developed and wealthy areas. But income distribution itself does not explain everything, since other regions of the State have more or less the same structure of income, or, in some of them, the earnings of the poor are even lower.14 In order to understand a model of urban development that continually reproduces new frontiers of precariousness, it is important to consider how land markets work as well as the impact of planning and land use regulations in shaping these markets. 3. Urban legislation and informal land markets in industrial cities the perverse links The case of Diadema, in the S~o Paulo metropolitan a region typies the process of large industrial expansion
The poorest area of the State is the Registro region (see map 13), comprised of 14 municipalities and 226,413 inhabitants, 45.17% of them considered indigent. The area around the capital, peripheral S~o a Paulo, consists of 38 municipalities and 6 million inhabitants, 5% of whom are considered indigent.
14

R. Rolnik / Geoforum 32 (2001) 471482 Table 1 Percentage of residences with an adequate situationa Group 1 City Batatais Barra Bonita Cerquilho Descalvado S~o Jos do Rio Preto a e Lins Itpolis a Penpolis a Araras Santa Rita do Passa Quatro Vargem Grande do Sul Socorro Ribeir~o Preto a Jaboticabal Rio Claro S~o Carlos a Amparo Catanduva Araraquara Bebedouro Mogi Guacu Bauru Santos Santa Rosa do Viterb Assis Pereira Barreto Marilia Botucatu Barretos Osvaldo Cruz S~o Joaquim da Barra a Limeira Campinas Fernad polis o Garca Americana Rio das Pedras Gua ra Franca Jales Piracicaba Santa Cruz do Rio Pardo Cruzeiro Cachoeira Paulista Mat~o a Leme Paraguacu Paulista Itatiba Santa Cruz da Palmeiras Amrico Brasiliense e Presidente Prudente Igaracu do Tiet^ e Rancharia Valinhos Santo Anastcio a Jundia Porto Feliz Vinhedo Morro Agudo Itu S~o Jos dos Campos a e Adequate (%) 74.00 73.39 71.61 71.58 70.72 70.13 69.87 69.50 68.84 68.14 67.70 67.62 67.61 67.51 67.50 67.14 66.51 66.46 65.71 65.58 65.43 65.22 65.12 64.61 64.42 64.32 64.28 64.08 63.96 63.87 63.64 63.12 62.67 62.07 62.06 62.02 61.60 61.56 61.40 61.30 61.30 61.08 61.03 60.95 60.80 60.34 59.93 59.50 58.95 58.88 58.62 57.93 57.41 57.11 56.83 56.66 56.42 56.19 55.99 55.88 55.87 Table 1 (continued) Group City Sorocaba Taubat e Lorena Indaiatuba Santa Barbara dest o Santo Andr e Cacapava Votuporanga S~o Paulo a S~o Bernardo do Campo a Braganca Paulista Conchal Taquarituba Pindamonhangaba Salto Andradina Santa Branca Jacare Votorantim Mogi das Cruzes S~o Roque a Itapeva Ribeir~o Pires a C^ndido Mota a 3 Salto de Pirapora Itupeva Po a Santa Isabel Mau a Campos de Jord~o a Atibaia Guarulhos Iguape Guaruj a Vrzea Paulista a Diadema Suzano Monte Mor Monguagu a Cajamar Franco da Rocha S~o Sebasti~o a a Caraguatatuba S~o Vicente a Santana do Parna ba Emb u Cotia Praia Grande Rio Grande da Serra Cubat~o a Francisco Morato Juquitiba Aruj a Embu-Guacu Adequate (%) 55.76 55.60 55.52 55.31 55.14 54.88 54.11 54.11 53.75 53.51 53.40 52.57 52.22 52.06 50.18 49.90 48.46 47.41 47.11 46.56 45.90 45.38 43.57 42.96 39.46 38.39 37.34 37.17 37.03 36.65 36.38 34.46 34.34 34.11 33.38 31.80 31.44 31.14 30.17 30.12 28.89 28.36 26.88 26.00 25.92 23.06 20.64 18.14 16.94 10.07 7.46 6.45 6.26 1.30

475

Source: IBGE (dierent years) Censo Demogrco 1991. a

in the metropolitan regions. When industry came to Diadema during the 1960s and 1970s, and in the case of Diadema, lasting until as late as 1990s, it accelerated demographic growth that generated a horizontal

476

R. Rolnik / Geoforum 32 (2001) 471482

expansion of great proportions, which was not really urban. Diadema was a town of 12,000 inhabitants in 1960. In 1971, a new highway linking the city of S~o Paulo to a Santos was built, crossing Diadema and attracting small and medium factories, which were satellites to large auto assembly plants in the municipality of S~o Bernardo. a Due to its location in the periphery of S~o Paulo, Diaa dema received only low-income migrants. In the absence of a signicant high-income group, only two land market segments were congured: one for industrial locations and another, for low income sub-divisions and housing. In 1973, a master plan for the city was approved and urban regulations were established, in which more that two-thirds of the city's territorial resources, the mosturbanized areas, were destined for industrial use. In addition to municipal regulations, in 1976, a state environmental law designed to protect the margins of metropolitan water reservoirs from urbanization excluded 724 ha or 23.55% of city territory from any urban use except very low-density, high-income housing, a market that simply did not exist in the city. The immediate eects of these regulations was an excess supply of land for industrial purposes (until 1990, approximately 40% of the total territorial resources destined for industrial use were empty), leaving a shortage of urban land destined for low income residential and other uses. This situation, in a context of high levels of demographic growth (20.42% per year in the 1960's and 11.23% in the 1970's) provoked an expansion along the city's outskirts, the occupation of non-urbanized areas, voraciously consuming all of the land which was not destined for industrial use, including the environmentally protected areas. With no other choice, since the land was either reserved for industrial use or very expensive due to limited supply, most of this expansion took place irregularly by informal land markets and did not meet any standards of urbanization.15 The result was that in the beginning of the 1980s, only 30% of the existing streets were paved, 50% of the households connected to water supply, 14% to sewerage and the infant mortality rate was 83/1000.16 This pattern has some similarities with the case of Guaruja, located in the periphery of the rapidly growing Santos region, where industrial plants (mainly in Cubat~o) and port facilities (Santos) were based.17 a Workers who did not have access to land and housing in
15

the urbanized portion of Santos, settled in the northwestern part of the neighboring island of Guaruja, forming the nucleus of Vicente de Carvalho, close to the channel that links the island to Santos. Like Diadema, in the 1960s and 1970s, the city of Guaruja attracted only low-income migrants, since management and technical employees of the region settled in the cities of Santos or S~o Paulo. a But Guaruj combines the ``bedroom community'' a with tourism, because its southeastern beach shore is a resort area ``A Prola do Atl^ntico ''.18 In this case, e a the strategy of the Guaruja Master Plan (approved in 1978) and urban regulations was to preserve the besturbanized land for tourism and simply ignored the creation of low-income residential markets, although the city was experiencing demographic growth.19 Therefore, the urbanized seashore that has always had water supply and sanitation, paved and lighted streets and steady investment in urban comforts, was microzoned for the dierent market segments for vacation homes, blocking access to low-income residential markets. This was accomplished by xing very detailed urban patterns that were completely unrelated to the economic and spatial logic of low-income settlements; i.e., a 500 m2 minimum lot for single family homes and multistory apartments. At the same time, low-income settlements for the permanent population expanded on Vicente de Carvalho over swamps and favelas were formed on hillsides overlooking the hinterland. With this pattern, it was possible to protect investment in weekend resorts, preventing the ``invasion'' of wealthy urbanized areas by the poor. What we have described, in both cases, is a very perverse mechanism that keeps poverty away from the best urbanized areas, establishing an invisible wall through urban regulations and reserving them for formal markets, while continually opening wild frontiers for informal markets. The mechanism is even more perverse given that the ``frontier'' opened to the informal markets was not occupied by formal markets, due to diculties in development or to environmental restrictions. The examples of Diadema and Guaruj, two munica ipalities with high rates of territorial exclusion demonstrate the failure of planning and urban regulation paradigms applied to the cities of S~o Paulo in the 1970s a and early 1980s. There is no connection between the presence of urban planning and the degree of territorial exclusion according to Fig. 1 which crosses the ranking of regulation (from the most to the least regulated city) with the
The rst development of the beach resort area took place in 1892, when the Companhia Balneria da Ilha de Santo Amaro established a a hotel, a casino and 50 chalets for the paulista coee elite. 19 The average annual population growth rate for Guaruj in the a 1970s was 5.26% a year.
18

From the total 380 opened parceling that city government have records today, 290 are irregular most of them being opened in the 70's (118). Source: Prefeitura Municipal (1995a). 16 Prefeitura Municipal de, 1995b. 17 Also some private container terminals, like Dow Chemical, Cargill and Cutrale were installed in Guaruj in the 70s, dening it also as a a retroport facility.

R. Rolnik / Geoforum 32 (2001) 471482

477

ranking of territorial exclusion (from the highest percentage of houses in adequate situation to the lowest). Contrary to conventional wisdom that the problem of Brazilian cities is a lack of planning, the research ndings illustrate that planning had very little impact on the social-environmental balance of the cities. In situations of high demographic pressure and competition for urban land, planning largely has been used as an instrument to ensure segregation and demarcation of markets. For that reason, planning needs to be profoundly reconsidered if it is to be able to confront the problems of this kind of urbanization the continuous production of territorial exclusion. Until now, we have not commented on the other extreme of the ranking the cities with the best performance in terms of providing adequate living conditions for their inhabitants. Although there is no city in the state with 100% of the households with adequate living conditions, the least precarious ones are located in the northeastern portion of the State,20 known as the ``California Paulista'' where ``agro-business'' has ourished.21 We will take the example of the Ribeir~o Preto a region, where the sugaralcohol complex was established in the 1970s, and where most cities have high rates of adequate living conditions, despite the fact that income concentration is about the same as that in S~o a Paulo and salaries are lower.

4. Agro-industrial ``closed'' cities Although the 1970s, when the ``Pro-Alcohol'' program was launched and consolidated,22 were years of economic expansion, the regional population growth rate of 2.45% per year was less than the state average for the same period (3.5% a year). During the 1980s, the eects of recession were much more acute in the metropolitan region and in the Baixada Santista than in Ribeir~o Preto. The region's population continued to a grow at rates (2.59%) slightly higher than the state average (2.02%). The region attracted migrants in the 1970s and 1980s, but never at the same level as the industrial centers and its peripheries. The explanation resides in the relationship between economic activity and urbanization. The sugaralcohol complex, as is the case with all agro-industrial sectors,
According to Table 1, best performances are found in Batatais 74%; Barra Bonita 73,39%; Cerquilho 71,61% and Descalvado 71,58%, all in agro-industrial regions. 21 The main crops are oranges and sugar cane, there is also cattle in the northern part of the region. 22 Pro-Alcohol is the Brazilian program to support alcohol production and consumption as a fuel for automobiles, which was created to face the oil crisis of the 1970s and 1980s, when strong dependency on imported oil led Brazil into a strong crisis after the international priceincrease of the product.
20

has its dynamic center outside of the urban areas. The logic of locating agro-industry is based on proximity to rural areas for farming raw materials rather unlike those of manufacturing economies. For this reason there is no concentration in a single polar city, but factories are spread out in dierent cities all surrounded by areas of production. Land for farming has appreciated in value a result of agro-industrial development, creating a barrier for converting it into urban space, even during cycles of economic and demographic expansion. The labor market is mostly seasonal strongest during the harvest since most of the farming has been mechanized and land has been concentrated in very large properties. During the harvest migrants settle in camps inside the plants sites, or sometimes in rented rooms, mostly in the smallest cities of the region. In order to avoid the permanent settling of these seasonal workers, cities establish control posts at highways, bus and train stations and sometimes blockades with checkpoints at the city entrances. The city of Jaboticabal was used as a case study. It is a medium-sized city in the region of Ribeir~o Preto, a a major agro-industrial center in the northeastern part of the State. With less demand for urban land due to its agro-industrial economy land prices are very low compared to Diadema or Guaruja. A portion of the income generated by agro-industrial production remains in the city, allowing for investment in urban infrastructure and creating less disparity in relative land prices (Table 2). In this case, the low-income population even considering low-wages and high-income concentration has better access to adequate housing. When the supply of urbanized land responds, for the most part, to demand, we do not see the phenomenon of super-appreciation of urbanized land; therefore, there is less territorial exclusion. However, this model has been supported only on a regional scale, since the poor who do not penetrate the city, settle elsewhere. Apart from that, the entire system is solely dependent on sugar cane production which has devastating environmental outcomes and is highly subsidized by the government. Thus, the model is not sustainable. 5. Territorial exclusion and violence in the state of S~o a Paulo Crime and violence in Brazil have moved to the forefront of attention for both policy-makers and the public (Caldeira, 1992). During the 1980s, mortality rates stemming from violence23 shifted from fourth to
23 In Public health records, the group encompasses all accidents, including trac accidents, suicides, homicides and other violence under the codes E800 to E999 from ChapterXVII of ICD (Souza, 1994).

478

R. Rolnik / Geoforum 32 (2001) 471482

Table 2 Relative land prices Guaruj, Diadema, Jaboticabal (1998)a a Price Average Most expensive Least expensive Dierence between two extremes
a

Diadema R$206,00 R$273,00 R$83,00 329

Guaruj a R$416,13 R$870,00 R$10,00 3700

Jaboticabal R$4100 R$100,00 R$1423 703

Sources: Diadema PMD Banco de Dados sobre o valor do preco da terra 1991/1998. Guaruj: PMG Planta de Valores Genricos 1998. a e Jaboticabal: prices research made in the city with brokers. Note: These relative land prices are not strictly comparable. In Diadema and Jaboticabal they reect with relative precision market prices, due to have been built upon market dates. In Guaruja, they have been built upon the Planta de Valores Genricos, the local database which denes land taxes gures for each parcel or building. Although the dates in Guaruja may have distortions e compared to market prices, they tend increasingly to converge with market prices.

second place in general mortality causes in the country. With a homicide rate of 23.35 per 100,000 in 1988,24 Brazil is the second most violent country in Latin America, the most violent region in the world. Only Colombia, with 89.5 homicides per 100,000 in the late 1980s is more violent (Ayres, 1998). With a homicide rate of 28.79 per 100,000 in 1991 and 29.70 for 1994, the S~o Paulo state is above the Brazilian average and is one a of the most violent states in Brazil.25 Much has been written about the relationship of crime and violence to illegal drug use and tracking. But it is dicult to generalize the assumption that the main cause of crime and violence is drug use and tracking, since it is a problem that aects each city dierently. It is most likely, that both an increase in violence and drug use and tracking are outcomes of the same causes. Traditional explanations associated violence with social marginalization and the inability of ruralurban migrants to make the transition from traditional to modern society. In the case of S~o Paulo, the argument a does not apply, since violence increased in the 1980s, at a time of a very sharp decline in ruralurban migration and when the state and particularly the metropolitan region experienced a demographic transition including declining population growth rates, fertility rates and shifting regional patterns of migration (Berqu, 1992). o As a matter of fact, 1991 Census data show that the city of S~o Paulo lost 900,000 inhabitants in the 1980s a while the greater metropolitan region had a migratory

Source Ministrio da Sa de and CENEPI, 1980/1989 Centro e u Nacional de Epidemiologia in Souza, 1994. 25 Data available for the S~o Paulo State for 1991/1994 (Fundac~o a a Seade, dierent years) is not available for all Brazilian States. However, comparative data on metropolitan regions and its capitals conrm S~o a Paulo's position. Although there is an increase in all the capitals of metropolitan regions, the homicide rate for S~o Paulo in 1988 was 38.9, a the highest rate, followed by Recife (37.8) (Souza, 1994) and Minayo and Souza (1993) point out that there is a problem with Rio de Janeiro records, due to a bias in the records about cause of death: if we take into account the number of homicides plus deaths registered as caused by rearms but whose intentionality or accidentality was ignored, the Rio de Janeiro homicide rate in 1988 climbs to 50.3.

24

inow of 450,000, most of these migrants moving from the nucleus to the periphery of the same region. (Ribeiro and Lago, 1995). Much of the recent literature makes a strong argument for the relationship of violence to poverty. The nexus is sustained by data showing that increases in violence occurred over a period where sharp increases in poverty levels also occurred. The recession of the 1980s and the eects of structural adjustment on the urban poor led to a severe drop in real wages and job opportunities for large segments of the labor force (Ayres, 1998). Although these assumptions are true for the case of S~o Paulo, absolute poverty itself a cannot explain why the S~o Paulo metropolitan region a is the most violent among Brazilian metropolises. In fact, the proportion of the population below the poverty line in S~o Paulo in 1989 (20.90%) is one of a the lowest among all Brazilian metropolitan regions. Salvador has 39.00% of the population below the poverty line and is one of the least violent metropolises of Brazil with a homicide rate of 17.5 per 100,000 (Singer, 1997). Surely the issue of inequality (or relative poverty) should be taken into account in order to explain higher crime rates, as some studies of North American cities point out (Freeman, 1996). But poverty, inequality and even changes in labor market structure can be dealt with in many dierent ways by households and individuals, depending on their vulnerability (Moser, 1996). According to Moser, the concept of vulnerability, or the insecurity of well being of individuals, households or communities in the face of a changing environment, captures not only the status of individuals and groups but also their means of resistance or the assets they can mobilize in the face of hardship. Our hypothesis here is that territorial exclusion makes individuals, households and communities particularly vulnerable, opening ground to violence and conict. The nexus between violence and territorial exclusion is very clear when we cross the data on the living conditions adequacy rate with Homicide rates for the 118 Sao Paulo cities surveyed (Table 3).

R. Rolnik / Geoforum 32 (2001) 471482 Table 3 Cities with the highest levels of territorial exclusion and homicide/100,000 rate, 1991 and 1994 Least adequate 1. Embu-Guacu 2. Aruja 3. Juquitiba 4. Francisco Morato 5. Cubatao 6. Rio Grande da Serra 7. Praia Grande 8. Cotia 9. Embu 10. Santana do Parnaiba 11. Sao Vicente 12. Caraguatatuba 13. Sao Sebastiao 14. Franco da Rocha 15. Cajamar 16. Monguagua 17. Monte Mor 18. Suzano 19. Diadema 20. Varzea Paulista 21. Guaruja 22. Iguape 23. Guarulhos 24. Atibaia 25. Campos de Jordao 26. Maua 27. Santa Isabel 28. Poa Adequacy (%) 1.3 6.26 6.45 7.46 10.07 16.94 18.14 20.64 23.06 25.92 26 26.88 28.36 28.89 30.12 30.17 31.14 31.44 31.8 33.38 34.11 34.34 34.46 36.38 36.65 37.03 37.17 37.34 1991 Homicide rate 44.95 18.85 50.68 58.34 37.42 30.33 44.95 36.89 78.84 10.86 20.19 45.61 50.68 25.91 44.78 26.62 15.82 25.94 85.58 19.1 26.34 14.38 44.12 15.17 21.69 58 15.87 39.52 Ranking 9 36 6 3 16 19 10 17 2 64 31 8 5 24 12 21 44 23 1 34 22 49 13 47 29 4 43 15 1994 Homicide rate 25.04 33.49 36.36 76.36 31.24 49.35 14.64 20.79 76.43 26.68 26.68 25.65 54.3 60.28 29.87 9.08 40.8 40.6 76.89 20.83 35.14 11.33 53.89 18.01 22.42 38.94 37.15 20.62 Ranking 29 19 16 3 20 7 48 34 2 27 26 28 5 4 25 75 10 11 1 33 18 58 6 41 30 14 15 36

479

Among the 28 worst situations of territorial exclusion (groups 3 and 4 in the cluster analysis in Table 1 from Embu-Guacu 1.30% of households in adequate condi tions to Po 39.52%), 25 have the worst violence india cators as well. The opposite is also true: among the 21 best situated cities in terms of adequacy, 14 are the least violent in the State. All other seven cities have homicide rates far below the State average (around 29 homicides per 100,000). Even centers known to be drug-tracking outposts Ribeir~o Preto and S~o Jos do Rio Preto a a e have homicide per capita rates lower than the State average (Table 4).26 Territorial exclusion makes daily life insecure and risky; as demonstrated throughout this paper. It blocks access to jobs, to educational and cultural opportunities, which are concentrated in small and protected enclaves within cities. It denies the possibility of using assets such as housing ownership to generate money and create jobs, since most of the houses are illegal and mixed use
26 REVISTA REVIDE (a local Ribeir~o Preto weekly magazine) a issued a special issue in 1998 on the raising violence in the city linked to drug tracking. The number of homicides grew from 83 in 1994 (the last year we have comparable data for all Sao Paulo municipalities) to 209 in 1997. Eight-ve percent of the victims had relations to drug trac.

of residences is generally forbidden by municipal land regulations. Excluded territories were created without the presence of Government or any ocial public sphere and thus were developed without any control or assistance. Public services, when they exist, are more precarious than in other parts of the city as assignment to these stigmatized territories is perceived by public employees as ``punishment''. Moreover, living under a permanent condition of denial of basic human needs makes inhabitants feel as if their lives are worthless. Recent studies of violence show that the main victims of homicide, the leading cause of mortality for youth, are young males, from 15 to 24 years (Cardia, 1998a,b).27 If we look at the demographic trends of those cities with the worst rates of violence, all of them had spectacular population growth rates in the 1970s and then began to decline in the 1980s: Diadema, Emb, Francisco Morato and others repeat this u pattern. These trends can be interpreted in terms of
The growing numbers of homicides among young males is an international trend. In the US, in 1987, at the peak of the crack consumption epidemics, 4223 young males between 15 and 24 years were murdered. In the same year, in Sao Paulo state alone 3171 young males were murdered (Cardia, 1998a,b).
27

480

R. Rolnik / Geoforum 32 (2001) 471482

Table 4 Cities with the lowest levels of territorial exclusion and homicide/100,000 rate, 1991 and 1994 Most adequate 1. Batatais 2. Barra Bonita 3. Cerquilho 4. Descalvado 5. Sao Jose do Rio Preto 6. Lins 7. Itapolis 8. Penapolis 9. Araras 10. Santa Rita do Passa Quatro 11. Vargem Grande do Sul 12. Socorro 13. Ribeirao Preto 14. Jaboticabal 15. Rio Claro 16. Sao Carlos 17. Amparo 18. Catanduva 19. Araraquara 20. Bebedouro 21. Mogi Guacu
a b

Adequacy (%) 74 73.39 71.61 71.58 70.72 70.13 69.87 69.5 68.84 68.14 67.7 67.62 67.61 67.51 67.5 67.14 66.51 66.46 65.71 65.58 65.43

1991 homicide rate 2.28 6.53 5.02 b 6.75 b 3.04 6.25 9.2 4.15 13.01 9.8 15.89 6.79 24.81 7.62 3.96 8.6 8.43 b 6.55

Ranking 2 21 15 24 5 19 38 8 54 41 70 25 85 28 7 34 32 35 23

1994 Homicide rate 2.2 a a 3.7 8.71 4.93 5.76 6.17 15.08 4.04 8.83 6.17 18.08 11.33 10.44 10.11 1.9 6.28 13.98 1.35 4.57

Ranking 3 8 32 13 16 17 67 9 35 18 74 55 50 47 2 21 62 1 10

No data available for 94. No data available for 1991.

generations: violence arises in the years following population growth. Young people, born in the 1970s, have lived their entire lives under territorial exclusion, and therefore, vulnerable conditions. By the end of the 1980s, when it came time for them to enter the labor market and form their families, there were no jobs and no means available to overcome the situation. Violence, with its ambiguous meaning of desperation and heroism, takes over. Research ndings provide evidence to link territorial exclusion and violence, and therefore shows their relationship with economic development models and planning policies. But this is just a map that indicates a special geography of urban conict in the contemporary Sao Paulo urban scene. In order to fully understand the causes of the problem and its eects on the lives of the people that live under violence, further research inside the communities is required. Questions about community perceptions of livelihood, security, the inuence of built environment, the role of police and justice and the availability of ``stocks'' of social capital can be understood only with a more participatory and anthropological approach. Without these perspectives, it will be very dicult to design policies to confront urban violence.28
28 For instance, an Americas Watch Committee report (1987) on violence in Sao Paulo indicates that 23.32% in 1982 and 14.9% in 1985 of registered homicides were committed by police corps, mostly the military police. (apud Adorno, 1997).

However, it is clear that territorial exclusion is one of the factors that opens ground to citizen insecurity and therefore, it is very important to address the following question:

6. How can urban policies face territorial exclusion? The traditional policy approach to low income settlements has been to ``forget'' or to stigmatize (labeling them as ``subnormal'') these territories in planning language and ``invest in them politically''. To invest politically means to negotiate, usually using votes in municipal elections as bargaining chips and providing investment in infrastructure and services as ``grants'' or ``favors'' from the mayor or city council representatives to the communities. This mechanism has been a very important source of local political power, since illegal or irregular settlements do not have the right to infrastructure and services as formal developments do. The illegal status of their environment makes low income inhabitants of informal settlements even more vulnerable to clientelism. The case of Guaruj is very representative of this a policy approach. Some illegal settlements were even promoted by city council members or spokesmen for the mayor, who distributed counterfeit rights to occupancy titles of non-urbanized land. In this case, political participation is organized vertically and lawlessness considered normal.

R. Rolnik / Geoforum 32 (2001) 471482

481

But the case of Diadema demonstrates that a dierent policy approach can lead to substantial changes in territorial exclusion rates. As we have seen, both, Diadema and Guaruj, grew a rapidly for over two decades and implemented a traditional planning strategy in the 1970s. In the two cases, housing markets for the poor were mostly informal. In the 1980s, however, the experiences began to dierentiate themselves; while in Diadema policies to overcome territorial exclusion were implemented, in Guaruja ocials continued with the traditional planning strategies, exacerbating exclusion. Diadema invested heavily in land regularization, massive implantation of infrastructure, in the improvement of conditions in favelas and changed its urban regulation strategy, introducing in its master plan instruments designed to amplify the supply of urban land for low income housing markets.29 The dierence between the two experiences and their results is mostly political. Diadema, due to its economic position had experienced organization through labor unions in the early 1980s. Beginning with a fund created by the families and friends of striking union members, a network of food cooperatives, housing associations and movements took root, generating an important source of social capital in the city. These organized groups began to demand city government intervention in urban conditions. In 1982, a local mayor was elected with the support of these networks and from that moment on, communities in Diadema became permanent interlocutors for urban policies, participating in the negotiations around regulation strategies and decisions related to city investment. In 1995, 121 favelas, among the existing 197 in Diadema, were urbanized. Sewerage covers 60% of households and water supply 95%. Ninety-six percent of the streets were paved and lighted, mortality rates dropped to 21/1000. These changes in the built environment give a chance for the next generation to have more assets to face economic diculties and to change their own perceptions of survival and self-respect. Caught within the same policy approaches and facing its rst crisis as a tourist resort, Guaruj gures were a worse: from the 17 favelas it had in 1980, there are now 57, with 47% of the population living in favelas; only 43,09% of the city is connected to water supply systems and 6.17% to sewerage.30 Territorial exclusion is thus, not a fatality or a theme that should be addressed only in economic terms. It remains to be seen, however, if this reduction in terriAmong the instruments used by Diadema, zonas especiais de interesse social (ZEIS) (special social interest zones) were created on land formerly destined to industry and loans were provided to housing associations to buy the land and build their houses. 30 Data from Prefeitura Municipal do Guaruj (1998). a
29

torial exclusion leads to a reduction in violence. According to Table 3, Diadema had in 1991 and 1994 the highest rates of violence. The politicians in charge in Diadema claim that their policies towards urban space will bring a reduction in violence, but we must take into account that the impact on the rate of violence will take at least one generation to be signicant. The youngsters that are killing and dying in Diadema today are the survivors of the high infant mortality rates of the early 80s when urban conditions were precarious. The case of Diadema, and others where territorial exclusion has been faced in Brazilian cities, may in the future provide us with clues which will demonstrate if investing in institution building and the commitment of local ocials to share power with low income communities can successfully transform risky urbanization towards a more equitable and sustainable model, with less danger of violent conict.

Acknowledgements This paper was written with the support of the Woodrow Wilson Center for Scholars, Washington, USA, as a part of the 21st Century Urban Agenda: Urbanization, Population, the Environment and Security Research Working Group, conducted at this institution in 199899 and was based in the research project Eects of the implementation of new land use controls on cities in S~o Paulo State, sponsored by the Fundac~o de Ama a paro  Pesquisa do Estado de S~o Paulo and the Lincoln a a Institute of Land Policy.

References
Adorno, S., 1997. La criminalidad violenta urbana en Brasil:tendencias y caracteristicas. In: Eunion sobre el desao de la violencia criminal urbana. Interamerican Development Bank, Rio de Janeiro. Auyero, J., 1999. This is a lot like Bronx, isn't it? Lived experiences of marginality in an argentine slum. International Journal of Urban and Regional Research 23 (1), 4569. Ayres, R.L., 1998. Crime and violence as development issues in Latin America and the Caribbean. In: Viewpoints. The World Bank Latin American and Caribbean Studies, Washington, DC. Berqu, E., 1992. 1991: um censo problematico. In: Novos estudos o Cebrap 3, Cebrap, S~o Paulo. a Caldeira, T.P.R., 1992. City of walls: crimes, segregation and citizenship in S~o Paulo. Ph.D. Dissertation on anthropology. a Graduate division of the University of California, Berkeley. Cardia, N., 1998a. A viol^ncia urbana e os jovens. In: Pinheiro, P.S. et e al. (Eds.), S~o Paulo sem medo um diagn stico da viol^ncia a o e urbana. Garamond, Rio de Janeiro. Cardia, N., 1998b. A paz na contra m~o da histria in Ser mdico. a o e Conselho Regional de Medicina do Estado de S~o Paulo, S~o a a Paulo. Castel, R., 1995. La piege de la exclusion in lien social et politiques. Revue Internacionale d'action Communautaire 34, Paris.

482

R. Rolnik / Geoforum 32 (2001) 471482 Rodgers, G., 1995. What is special about a `social exclusion' approach? In: Rodgers, G. et al. (Eds.), Social Exclusion: Rhetoric, Reality, Responses. International Institute for Labour Studies/United Nations Development Programme, Geneve. Singer, P., 1997. Social exclusion in Brazil. International Institute for Labor Studies/Labor Institutions and Development Programme, Geneve. Souza, E.R., 1994. Homic dios no Brasil: O grande vil~o da sade a u p blica na dcada de 80. In: Cadernos de sade pblica, vol. 10. u e u u Fundac~o Oswaldo Cruz, Rio de Janeiro. a Sposati, A. et al. 1996. Mapa da exclus~o social da cidade de S~o a a Paulo. S~o Paulo: EDUC. a Telles, E., 1995. Structural sources of socioeconomic segregation in Brazilian metropolitan areas. American Journal of Sociology 100 (5). Raquel Rolnik is an Architect and Urbanist, Ph.D. in Urban History New York University, a Researcher in Plis Institute S~o Paulo, is o a Coordinator of the Master Program in Urbanism Pontif Univercia sidade Catlica de Campinas. o

Freeman, R., 1996. Why do so many young American men commit crimes and what might we do about it?. Journal of Economic Perspectives 10 (1). Fundac~o Seade, dierent years. Anurio estatistico do estado de S~o a a a Paulo. Fundac~o SEADE, S~o Paulo. a a Minayo, M.C., Souza, E.R., 1993. Viol^ncia para todos. In: Cadernos e de sade publica, vol. 9. Fundac~o Oswaldo Cruz, Rio de Janeiro. u a Ministrio da Sa de and CENEPI, 1980/1989. Centro Nacional de e u Epidemiologia. Subsistema de informac~o sobre mortalidade. a Ministrio da Sade, Bras e u lia. Moser, C., 1996. Confronting crisis a comparative study of household responses to poverty and vulnerability in four poor urban communities In: Environmentally Sustainable Development Studies and Monographs Series 8. The World Bank, Washington. Paugam, S. (Ed.), 1996. L'exclusion: l'etat des savoirs. Editions La Decouverte, Paris. Prefeitura Municipal de Diadema, 1995a. Cadastro e banco de dados. Secretaria de Habitac~o e Desenvolvimento Urbano, Diadema. a Prefeitura Municipal de Diadema 1995b. Sumrio de dados scioa o econ^ micos. Scretaria de Habitac~o e Desenvolvimento Urbano, o a Diadema. Ribeiro, L.C.Q., Lago, L.C., 1995. Din^mica metropolitana e novos a padr~es de desigualdade social in S~o Paulo em Perspectiva 9:2. o a a Fundac~o Seade, S~o Paulo. a