Variáveis que influenciam a formação de preços do Café Arábica: uma análise regional e nacional

• Ano de defesa: 2020
• Autor(a) principal: Vilela, Eunice Henriques Pereira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Administração
• Palavras-chave em Português: Café; Coffee; Regionalidade; Regionality; Custos de Produção; Production Costs; Redes Neurais Artificiais; Artificial Neural Networks; Gestão de Riscos; Risk Management; CNPQ::CIENCIAS SOCIAIS APLICADAS
• Link de acesso: https://repositorio.ufu.br/handle/123456789/28693; http://doi.org/10.14393/ufu.di.2020.4

The coffee production is an activity of great economic and social relevance for Brazil and especially for the state of Minas Gerais, the main producing state and responsible for more than half of national production. Considering these characteristics and the importance of production cost and risk management for the profitability and sustainability of the sector, this dissertation is divided into two chapters. The first chapter focus on coffee growing in the state of Minas Gerais, with the objective of analyzing the behavior of production costs in relation to the price of coffee in the main producing regions of the state, namely: the South of Minas, the Cerrado Mineiro and the Matas de Mines; besides verifying the influence of the region as a differentiating factor on the behavior of these variables. For this, we used a panel data regression that had as dependent variable the price paid to the producer for the Arabica coffee bag in the aforementioned regions, between 2007 and 2018, and the Kruskal-Wallis test, to identify possible relationships between cost variables and the producing region. The results indicated that machine costs, pesticides and volume produced have a negative relationship with coffee price changes, while taxes have a positive relationship, and that the region has a significant relationship with coffee price changes in coffee. producing regions of Minas Gerais. It was also identified that productivity, costs with pesticides, labor and machines present different distributions of values between regions. The second chapter, in turn, aimed to elaborate and validate a forecasting model for the price behavior of Brazilian coffee through the use of Artificial Neural Networks. To compose the forecast model, variables related to coffee production costs were used in the main producing municipalities of Brazil, macroeconomic variables that may affect the prices of agricultural commodities, variables related to the Brazilian coffee market and variables related to the world coffee market. Models were built and trained with twelve arrangements of variables, which differed by the variables that made up their inputs. With an MSE of 0.0958 and an R² 0.8394, the model with the best predictive performance was the arrangement composed of production cost variables such as machinery, labor, fertilizers, pesticides and financial expenses, and interest rates. , foreign exchange, taxes, and for the consumption, export, stocks and production of coffee in Brazil, in the configuration with 13 neurons in the first hidden layer and 1 neuron in the second hidden layer.