TIME SERIES ANALYSIS OF THE BEHAVIOR OF NATURAL RUBBER PRICES IN THE INTERNATIONAL MARKET

This work analyzed the behavior of natural rubber prices in the international market from January 1982 to December 2006 in function of its aggregated demand and supply, pointing out the main producing and consuming countries. Specifically, the research studied the evolution of prices and of the marketed quantum of natural rubber in the international market. It was characterized, identified, estimated and analyzed models for the real monthly prices series of raw rubber RSS 1 (US$/t), and the accuracy of the estimated models for forecasting prices of this commodity was tested from Jan/2006 to Dez/2006. The studied models were of ARIMA-ARCH class. The main results were: the real natural rubber prices presented decreasing tendency in the period being studied; the ARIMA family estimated model indicating the existence of heteroskedasticity in the series, making it necessary to identify, to estimate and to analyze the models of ARCH family; the model which best adjusted the returns of the price series of the raw rubber RSS1 was AR(1)-GARCH(1,1); the models of the ARIMA family didn't satisfy the prognosis conditions of the series being studied; the AIR (1)-GARCH (1,1) model was accurate for forecasting rubber prices.