Indicadores de renda baseados em consumo de energia elétrica: abordagens domiciliar e regional na perspectiva da estatística espacial

• Ano de defesa: 2010
• Autor(a) principal: Francisco, Eduardo de Rezende
• Orientador(a): Aranha Filho, Francisco José Espósito
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Renda domiciliar; Consumo de energia elétrica; Classificação econômica; Estatística espacial; Pesquisa domiciliar
• Palavras-chave em Inglês: Household income; Electricity consumption; Economic status; Spatial statistics; Residential survey
• Link de acesso: https://hdl.handle.net/10438/8158

In order to evaluate the use of Electricity Consumption as a Socioeconomic Status, this research analyzes information in two levels of geographical aggregation. At the first level, under a territorial perspective, it investigates indicators of Income and Electric Energy Consumption aggregated by weighted areas (set of census sectors) in the city of São Paulo and uses the microdata of Demographic Census 2000 jointly with residential consumers’ database of AES Eletropaulo. It applies Spatial Auto-Regressive (SAR) models, Geographically Weighted Regression (GWR), and an unprecedented combined model (GWR+SAR), developed in this study. Several neighborhood matrices were used to assess the influence of space (with Downtown-Suburbs pattern) of the variables under study. The variables showed strong spatial autocorrelation (Moran's I greater than 58% for the Energy Consumption and more than 75% for the Household Income). Relations between Income and Electricity Consumption were very strong (coefficients of determination of Income reached values from 0.93 to 0.98). At the second level, the household one, it uses data collected in the Annual Satisfaction Survey of Residential Customer, coordinated by the Brazilian Electricity Distributors Association (ABRADEE) for the years 2004, 2006, 2007, 2008 and 2009. Weighted Linear Model (WLM), GWR and SAR were applied to survey data with interviews allocated on the centroid and the seat of the districts. For the year 2009, we obtained the actual locations of the households interviewed. Additionally, 6 algorithms of points distribution within the polygons of the districts have been developed. The results from models based on centroids and seats obtained a coefficient of determination R 2 of around 0.45 for the GWR technique, while the models based on scattering points within the polygons of the districts have reduced this account to about 0.40. These results suggest that the algorithms of allocation of points in polygons allow the observation of a more realistic association between the constructs analyzed. The combined use of the findings shows that the billing information of the electricity distributors has great potential to support strategic decisions. Because they are current, available and monthly updated, socioeconomic indicators based on energy consumption can be very useful as an aid to processes of classification, concentration and estimation of household income.