Inequalities in Space and Time: Population Segregation and Urban - PDF document

Inequalities in Space and Time: Population Segregation and Urban Morphology in Brazil, 2000/2010 1 Igor Cavallini Johansen 2 Roberto Luiz do Carmo 3 Abstract The Urban Transition, that is, the passage from a predominantly rural to a mostly urban society, occurred in Latin America faster compared to what happened in Western European and North American countries. In Brazil, the rapid urban growth produced cities marked by social inequality. Focusing on this country, our purposes on this study were: 1) to investigate how social segmentation occurs in the urban fabric or, in other words, how social inequality is reflected in the city space; 2) check if there are any changes in progress, i.e., if that pattern is being modified; and 3) with regard to urban morphology, test the hypothesis that the center-periphery pattern remains valid to explain the configuration of urban space in Brazilian metropolises. This analysis was performed using the five largest metropolitan regions of Brazil according to their total population size in 2010. This study investigated the years 2000 and 2010, using spatial analysis and statistics tools. It was concluded that, in Brazil, despite a small reduction in socio-spatial segregation between 2000 and 2010, this topic continues being a relevant issue to the population research agenda for the next decades. Key-words: Segregation; Inequality; Spatial Analysis; Urban Morphology; Brazilian Metropolises. 1 Paper presented in the XXVIII IUSSP International Population Conference – Cape Town, South Africa, October 29 th to November 04 th 2017. 2 Universidade Estadual de Campinas, Instituto de Filosofia e Ciências Humanas, Programa de Pós-Graduação em Demografia. Campinas/SP, Brasil. Contato: igor@nepo.unicamp.br. 3 Universidade Estadual de Campinas, Instituto de Filosofia e Ciências Humanas, Programa de Pós-Graduação em Demografia. Campinas/SP, Brasil. Contato: roberto@nepo.unicamp.br. 1

Introduction Latin American countries experienced a rapid urbanization process that occurred more explicitly during the second half of the 20th century (GWYNNE, 1985; FARIA, 1991; OJIMA, 2007; MARTINE, MCGRANAHAN, 2010). The Urban Transition, that is, the passage from a predominantly rural to a mostly urban society, occurred here faster compared to what happened in Western European and North American countries (UN, 2014). The rapid growth of cities, especially considering the substantial influx of populations from rural to urban areas, culminated in the production of cities marked by social inequality. The segmentation of social groups by different income strata reflected in the infrastructural differences of the dwelling place (KOWARICK, 1979; MARICATO, 2000). Particularly in Brazil, the peripheries, understood as areas or regions not adequately provided with basic services of urban infrastructure (COSTA, MONTE-MOR, 2002), took increasing space in cities, what was even more evident in the metropolitan regions (KOWARICK, 1979; MARICATO, 1979; LAGO, 2000). Thus one of the clearest marks of Brazilian urbanization is the strong division between social groups. Consequently, the segments with the worst socioeconomic conditions, whose dwelling places have basic deficiencies in access to urban resources and services, have in these places a reaffirmation of their precarious living conditions and, consequently, the maintenance of the poverty cycle (MARICATO, 1979; 2000; TASCHNER, BOGUS, 2001). The purposes of this study were: 1) to investigate how social segmentation occurs in the urban fabric or, in other words, how social inequality is reflected in the city space; 2) check if there are any changes in progress, i.e., if that pattern is being modified; and 3) with regard to urban morphology, test the hypothesis that the center-periphery pattern remains valid to explain the configuration of urban space in Brazilian metropolises. The center-periphery model suggests the organization of the population groups in concentric circles according to their socioeconomic conditions (BURGESS, 1929; TASCHNER; BOGUS, 2001). This explanatory model predicts the occurrence of a "rich" center versus a "poor" periphery. In the center there would be areas better equipped with urban resources and services and, consequently, they would be inhabited by the most affluent population segments. In the surrounding areas, in turn, marked by the precariousness of these resources and services, would be located the population groups in worse socioeconomic conditions. These population groups would reside there because they cannot afford living in the more central areas, where the land price is considerably higher (SINGER, 1978; OLIVEIRA, 1979; MARTINE, MCGRANAHAN, 2010). 2

Methods Study areas The five largest Brazilian Metropolitan Regions (MR) were selected in terms of population size in 2010. They were, in descending order: São Paulo MR, Rio de Janeiro MR, Belo Horizonte MR, Porto Alegre MR and Integrated Development Region of the Federal District and Surroundings (whose abbreviation is named RIDE-DF in Portuguese and where the country capital is located – Brasília). They belong to the South, Southeast and Center-West regions of the country, guaranteeing the coverage and geographical representativity of the Brazilian metropolises. Data Source The analysis used as data source the Atlas of Human Development in the Brazilian Metropolitan Regions. The Atlas was launched in 2014 and was developed by the United Nations Development Program (UNDP), the Institute of Applied Economic Research (IPEA) and the João Pinheiro Foundation (FJP). Its data refer to the years 2000 and 2010, while based on information from the Brazilian Population Census carried out in these years. Unit of Analysis The unit of analysis of this investigation is the Human Development Unit, produced by the Atlas of Human Development in the Brazilian Metropolitan Regions. Variable The Human Development Index (HDI) is a composite indicator that synthesizes information of income, schooling and life expectancy. It is regularly used by the United Nations to assess the level of development of the countries (UN, 2016). In this study it is used the HDI calculated for the intramunicipal level, available in the Atlas. Population Projection Considering that some units of analysis had no population in 2000 and, therefore, had no value for the variable HDI, it was necessary to carry out the projection of the missing data. The units were not expressive in the total (RMSP: 4.6%, RMRJ: 2.9%, RMBH: 2.9%, Porto Alegre RM: 3.6%, RIDE- DF: 3.3%). These gaps were adequately met using the Forecast function of Microsoft Excel 2010 software. 3

Spatial statistical analysis The processes of the spatial statistics analysis as well as the representation of the results on maps were carried out from ArcMap software version 10.3.1. The statistical analysis used three concomitant and complementary procedures: 1) Spatial Autocorrelation (Moran ’ s I): This tool is also known as the Global Moran’s I. It is used to verify if there are groups of units of analysis (also called clusters) with high IDHM on the one hand, and low IDHM on the other. To reject the null hypothesis (H0), it is used as criterion of statistical significance p-value <0.01 4 2) Incremental Spatial Autocorrelation: this is a strategy to find the best radius distance (in meters) that the clusters should be searched in order to boost the results. The choice is made by numerous tests performed by the software in order to find the distance that points the best z-score possible. The z-score comprises a measure that assists in the analysis of the existence of clusters. The ideal distances found in the analysis for 2010 were also applied for 2000 (see next step), so that the results could be compared between the two years. 3) Cluster and Outlier Analysis (Local Moran ’ s I): this comprises the Moran Local Index. In this step, we can use as input the best radius distance found in the previous one of the analysis. The result will be a map, which classifies the units of analysis into five categories: Not Significant – Area units where the analysis was not significant; High-High – Area units with high HDI close to others with high HDI; High-Low – Area units with high HDI close to others with low HDI; Low-High – Area units with low HDI close to others with high HDI; Low-Low – Area units with low HDI close to others with low HDI. Results The metropolises selected for this study present differences between themselves, both in terms of expansion history (which will not be addressed in this study) and about their current demographic characteristics (Table 1). The Brazilian metropolis that occupies the first place in population volume is the São Paulo MR, which had in 2010 about 19.7 million inhabitants, while the fifth largest is the RIDE-DF, with more than 3.5 million population. The geometric rate of annual growth of the population between 2000 and 2010 also varies, being the greater 2.3% per year for the RIDE-DF and the smallest, 0.8% for Rio de Janeiro and Porto Alegre. The degree of urbanization among the surveyed metropolises is between 94.1% and 99.8% 4 In the case of the Global Moran’s I, the null hypothesis (H0) is that there would be no clusters. The aim is to reject the null hypothesis. Therefore, in the result of the analysis, the p-value, as indicated, should be less than 0.01. 4