Análise dos preços de carvão vegetal em quatro regiões no estado de Minas Gerais

• Autor(a) principal: Rezende, José Luiz Pereira de
• Data de Publicação: 2015
• Outros Autores: Coelho Junior, Luiz Moreira, Oliveira, Antônio Donizette de, Sáfadi, Thelma
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Repositório Institucional da UFLA
• Texto Completo: http://repositorio.ufla.br/jspui/handle/1/14240
• topic: Carvão vegetal; Economia florestal; Série temporal; Tendência; Sazonalidade; Charcoal; Forestry economy; Time series; Tendency; Seasonality

The State of Minas Gerais is the largest producer and consumer of charcoal, that is used as term-reducer of iron ore, for producing pig iron. This study analyzed the time series of charcoal prices in four regions of Minas Gerais State. For the analysis of the price series, the SARIMA model was used, for finding a model that better forecasts prices for the four studied areas. The most appropriate models were chosen using graphical analyses of the standardized residues, autocorrelation functions and partial autocorrelations, stochastic tests and criteria of evaluation of the order of the model. It concluded that: the differences of charcoal prices occur, basically, due to the geographical location; the analyses of domain of the time and domain of the frequency showed that there is difference in the price series of the four studied areas; the areas of Sete Lagoas and Belo Horizonte, giving that they are closely located, possess similar prices and they generated similar model; the studied areas presented differentiated models and supplied good adjustments for the observed series. The best models were SARIMA (2,1,1)x(1,0,0)12, for Belo Horizonte; SARIMA (2,0,0)x(2,1,2)12, for Divinópolis; SARIMA (2,1,1)x(1,0,0)12, for Sete Lagoas and SARIMA (1,1,1)x(1,1,1)12, for Vertentes. Such models presented in a parsimonious way, containing a small number of parameters. All models SARIMA (p,d,q) (P,D,Q)s, for the four studied areas, presented white noise and supplied adequate price forecast.