Estudo do mercado de fertilizantes no Brasil por meio de previsões estatísticas

Abstract

Due to population growth and rising demand for food, there is a trend towards increased use of fertilizers in agriculture. In this sense, the objective of the present work was to perform a study of the NPK fertilizer market in Brazil, from mathematical modeling to obtain future projections, as well as to evaluate and compare the methods used for forecasting. Firstly, the current state of the global and Brazilian fertilizer market was discussed and concepts about quantitative methods of forecasting and evaluation criteria were presented. After that, historical data were collected regarding the production, consumption, exports and national imports of fertilizers in a reliable database, such as the National Association for the Dissemination of Fertilizers (ANDA), the United Nations Organization for Agriculture and Food (FAO) and the Brazilian Institute of Geography and Statistics (IBGE). From this collection, the forecasts were elaborated using the methods of moving averages, exponential smoothing, ARIMA and multiple linear regression, using the Microsoft® Excel, Action Stat®, NNQ - Statistics and gretl® programs. For the variables imports, exports and fertilizer consumption, the most suitable model for predictions was the one of exponential smoothing, whereas for the production variable the ARIMA method was obtained as best result. In relation to multiple linear regression, a model was found to explain fertilizer consumption in Brazil from the determinants of planted area and agricultural production. All projections were satisfactory, as they did not exceed the upper and lower forecast limits.