%0 Journal Article %T FORECASTING RESIDENTIAL ELECTRICITY CONSUMPTION IN BRAZIL: APPLICATION OF THE ARX MODEL PREVIS O DO CONSUMO RESIDENCIAL DE ENERGIA EL¨¦TRICA NO BRASIL: APLICA O DO MODELO ARX %A Joao Bosco de Castro %A Alessandra de ¨¢vila Montini %J Future Studies Research Journal : Trends and Strategies %D 2011 %I Funda??o Instituto de Administra??o %R 10.7444/future.v2i2.50 %X This work aims to propose the application of the ARX model to forecast residential electricity consumption in Brazil. Such estimates are critical for decision making in the energy sector, from a technical, economic and environmentally sustainable standpoint. The demand for electricity follows a multiplicative model based on economic theory and involves four explanatory variables: the cost of residential electricity, the actual average income, the inflation of domestic utilities and the electricity consumption. The coefficients of the electricity consumption equation were determined using the ARX model, which considers the influence of exogenous variables to estimate the dependent variable and employs an autoregression process for residual modeling to improve the explanatory power. The resulting model has a determination coefficient of 95.4 percent and all estimated coefficients were significant at the 0.10 descriptive level. Residential electricity consumption estimates were also determined for January and February 2010 within the 95 percent confidence interval, which included the actual consumption figures observed. The proposed model has been shown to be useful for estimating residential electricity consumption in Brazil. Key-words: Time series. Electricity consumption. ARX modeling. O objetivo deste trabalho ¨¦ propor a aplica o do modelo ARX para fazer previs es do consumo residencial de energia el¨¦trica no Brasil, fundamentais para suportar decis es do setor de energia em bases t¨¦cnica, econ mica e ambientalmente sustent¨¢veis. A fun o da demanda de energia el¨¦trica foi modelada segundo um modelo multiplicativo com base na teoria econ mica e incluiu quatro vari¨¢veis explicativas: tarifa residencial de energia el¨¦trica, rendimento m¨¦dio real das pessoas ocupadas, ¨ªndice de infla o de utilidades dom¨¦sticas e consumo de energia. Para estimar os coeficientes da equa o do consumo de energia el¨¦trica, foi utilizado o modelo ARX, o qual considera na proje o da vari¨¢vel dependente a influ¨ºncia de vari¨¢veis ex¨®genas e modela os res¨ªduos por meio de um processo auto-regressivo a fim de aumentar seu poder explicativo. Resultados indicaram um coeficiente de determina o de 95,4% e todos os coeficientes estimados foram estatisticamente significantes a um n¨ªvel descritivo de 0,10. Foram realizadas proje es do consumo residencial de energia el¨¦trica para os meses de janeiro e fevereiro de 2010 no intervalo de c %U http://www.revistafuture.org/index.php/FSRJ/article/view/50