Aplicação de singular spectrum analysis para a exportação do minério de ferro brasileiro

Abstract

The iron ore sector plays a central role in the Brazilian economy, especially in relation to the trade balance. In 2020 the sector was responsible for 12.3% of Brazil's total exports, thus assuming the second position in the ranking of the country's exports. Since the 2000s, due to the accelerated process of urbanization and industrialization in China, there has been a significant growth in demand for iron ore. Thus, given the importance of the sector for the country, coupled with uncertainties about the perpetuation of Chinese demand, predictive models for iron ore exports are relevant. Thus, this paper aims to model the iron ore export séries from Brazil and the states of Minas Gerais and Pará using the Singular Spectrum Analysis (SSA) methodology and compare its predictive ability with the classical Exponential Damping (AE) and ARIMA methods. The results of the adherence statistics show that the forecasts made by SSA present the best results for all the analyzed séries. Thus, this work contributes by expanding the studies of forecasting in one of the most relevant sectors in the country.